i am trying to make a peltier powered mini fridge and i am having some trouble getting it to get to the temperatures i want.

First of all, some info. I am using a Cooler Master T4 as the cooler for the peltier's hot side. The problem is that the contact surface of the fan's heatsink is smaller than the peltier's, so i get really hot edges. That is why i bought a 1mm thick copper sheet to put between the peltier and the fan.

I am powering it all with an old 350W computer power supply. It should be enough.

I first bought a 12706 peltier element and tried the setup. Since i don't have an accurate thermometer, i meassured the performance by measuring the current that the peltier drawed and checked if the cold side was cold enough to freeze water drops.

If i don't put the copper sheet, and just add some thermal paste, i get around 3.8 A and the cool side is cold enough to freeze water. My concern with this setup is that the edges of the peltier(that are not in contact with the heatsink) are really hot and i am afraid to damage the element.

If i add the copper and put some thermal paste in both sides, i get 2.5A, that is much less than the previous setup. This way i can't get it cold enough.

Then i bought two more peltier modules, the 12710 this time, and some thermal pad to repalce the thermal paste for something more durable.

This time i get around 5.4A when using just the thermal paste and 5.2A with just the thermal pad. The edges are still really hot, but at least i am able to freeze water. It doesn't seem to be any colder that the 12706 though.

However i am noticing that the wires in the peltier get hot, so i am suspecting that they are limiting the amount of current. Maybe that is why i don't get best performance with the 12710.

So, what am i doing wrong? I know for example that the copper sheet should be as flat as possible, maybe that is why i don't get proper thermal conduction.

Should the 12710 get colder than the 12706? Because if not i am wasting a lot of power for nothing.

Will my peltier get damaged if i let it run with hot edges?


  • \$\begingroup\$ It's a waste of time to run it with hot edges, get a larger hot side heatsink, or a better heatspreader if you want to stay with the T4. Don't use thin copper sheet, unlilkey to maintain the flatness, use much thicker alli plate, with ground faces. Or, dead PCs can be collected for free from dumps, rip a bigger CPU heatfansink off one. \$\endgroup\$ – Neil_UK Nov 6 '16 at 20:58
  • \$\begingroup\$ The current will go down as the temperature difference increases- you may need more voltage as well as a much better hot-side heatsink. \$\endgroup\$ – Spehro Pefhany Nov 6 '16 at 23:11
  • \$\begingroup\$ If one reads between the lines OP calls the Fan and Heatsink combination just Fan and this may be confusing the members here. I think the OP interspersed a copper heat spreader to match the Peltier and the Fan and Heatsink dimensions. \$\endgroup\$ – KalleMP Jan 22 '18 at 18:41

As Neil_UK has mentioned, a 1 mm copper sheet is not going to do you any good. You need a much beefier heatsink, and it needs as many fins as you can get. If you try your copper again, you'll notice that the copper is getting very hot. This is why the cold side is not getting cold. The TEC is inefficient with large temperature differences between the two sides. So you need to keep the hot side as cool as possible in order to keep the cold side really cold. You do this by making it as easy as possible for the heatsink to warm up something else, like the air being blown across it but the fan, and for this the more surface area the better. So, thick copper and lots of fins.

Oh yes, and about thermal paste. The coating must be as thin as possible while still filling the gap between the TEC and the heatsink. Don't think you can slather on a good layer and then clamp the heatsink on. The thermal compound just isn't that good a heat conductor, but by filling in all the minute pits, scratches and voids caused by the two surfaces not being perfect it can markedly improve the transfer of heat to the heatsink. But too much is worse than too little.

For a thin heatsink, in addition to its poor intrinsic properties, you must be aware that clamping it will inevitably distort it, often pulling it away from the Peltier in the process. That's another reason for using thick heatsinks, and as Neil stated, they must be precision machined to be very flat to match the TEC.

  • \$\begingroup\$ I like to add: If there isn't a good reason to transport heat away by conduction, do it, by any means, by convection. Even a non-optimal heatpipe arrangement is at minimum 10 times better than anything utilizing peltier elements. \$\endgroup\$ – Janka Nov 6 '16 at 22:30
  • \$\begingroup\$ And sometimes we find heat-pipes used to take the heat away from the Peltier element to the large air cooled fin area. A lot of high power CPU heat-sinks use heat-pipes these days. \$\endgroup\$ – KalleMP Jan 22 '18 at 18:37
  • \$\begingroup\$ It sounds like the 1mm sheet is just a shim to transfer the heat between the Peltier element and the real heatsink as they don't line up otherwise. He's not trying to dissipate all the heat with a 1mm copper sheet. \$\endgroup\$ – immibis Jun 26 '18 at 3:13
  • \$\begingroup\$ @immibis - Nope. "That is why i bought a 1mm thick copper sheet to put between the peltier and the fan" No mention of heatsink. \$\endgroup\$ – WhatRoughBeast Jun 26 '18 at 14:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.